The first cases of COVID-19 comes from China (Wuhan) and were discovered in November 2019. From the beginning of the 2020 new outbreaks of the coronavirus have started to appear in other countries. The pandemic spread around the world rapidly. The discovery of the easily contagious virus forced on the world’s government to take an fast and unconventional actions. Many countries have closed their borders and introduced lockdown. Due to such actions many economies suffer from a pandemic.
The aim of this report is to discover whether the COVID-19 had an impact on unemployment. Firsly, it will be examined from the point of view of OECD countries, G7 countries, the USA. After that, the situation in Poland will be presented.
Dataset sources:
- Wikipedia information regarding coronavirus cases.
- Monthly unemployment rate in OECD countries comes from the OECD website.
- Poland unemployment rate in voivodships comes from Poland’s Central Statistical Office.
- Poland map was built from dataset introduced during the classes.
Although, the first cases of coronavirus were discover in China in the late of 2019, the pandemic rapidly spread around the whole world. Currently, there is 174,432,190 coronavirus cases, 3,758,217 people died (10 June, 2021 update).
Official top 3 countries with highest COVID-19 cases number:
- United States (33,566,719 cases),
- India (29,089,069 cases),
- Brazil (17,038,260 cases).
Below graphs present the situation of unemployment rate in OECD countries. It means that the unemployment rate is captured in 37 countries. The first lollipop graph shows 15 highest unemployment rates among OECD countries in 2020-03. Country with the highest unemployment rate is GRC (around 15.9%), then CRI (around 15.7%) and ESP (around 14.5%).
The second lollipop graph shows 15 largest unemployment rate in 2021-04. During that 1 year it can be observed that on average unemployment rate decreased in almost every country.
It needs to be emphasize that there are some countries that have a problem with high unemployment rate independently from pandemic. Due to that, one can find the barchart more informative for that kind of analysis. From that graph one can capture that the highest change between 2020-03 and 2021-03 is observed in ITA, then SWE and ISL.
The aim of this interactive graph is to capture the situation in G7 countries. It can be an important analysis due to the nature of these countries. The Group of Seven countries represent a large part of the global economy.
Presented type of graph (Plotly) is usefull as it is easy and fast to capture the country (or countries) of interest as well as time horizon.
In case of the coronavirus impact on the unemployment rate the most important are the dates related to 2020 and 2021.
In our opinion the most interesting situation is in United States, as it is a country with highest COVID-19 number of cases.
The below graph can easily reflect the situation of Poland in comparison to the European Union countries during timeframe from 2021-03 to 2020-03.
During the observed time the lowest unemployment rate in European Union was equal 6.7%, the highest was equal 7.8%. In case of Poland, it was 3.1% and 3.4%, respectively. Polish situation in comparison to European Union countries seems better as unemployment rate is lower.
One can chose any country and get country-specific report
We also did some forecasting using ARIMA. We used function auto-arima to find the best parameters.
## [,1]
## 2000-12-01 16.8
## 2001-01-01 17.1
## 2001-02-01 17.4
## 2001-03-01 17.7
## 2001-04-01 18.1
## 2001-05-01 18.4
## [1] 0.01192257
## [1] "ok"
##
## Fitting models using approximations to speed things up...
##
## ARIMA(2,1,2)(1,0,1)[12] with drift : -312.6321
## ARIMA(0,1,0) with drift : -108.7181
## ARIMA(1,1,0)(1,0,0)[12] with drift : -307.8759
## ARIMA(0,1,1)(0,0,1)[12] with drift : -226.1317
## ARIMA(0,1,0) : -90.48257
## ARIMA(2,1,2)(0,0,1)[12] with drift : -309.2965
## ARIMA(2,1,2)(1,0,0)[12] with drift : -312.3843
## ARIMA(2,1,2)(2,0,1)[12] with drift : -333.4897
## ARIMA(2,1,2)(2,0,0)[12] with drift : -335.5864
## ARIMA(1,1,2)(2,0,0)[12] with drift : -333.1777
## ARIMA(2,1,1)(2,0,0)[12] with drift : -337.1392
## ARIMA(2,1,1)(1,0,0)[12] with drift : -313.1164
## ARIMA(2,1,1)(2,0,1)[12] with drift : -335.108
## ARIMA(2,1,1)(1,0,1)[12] with drift : -314.1705
## ARIMA(1,1,1)(2,0,0)[12] with drift : -332.2236
## ARIMA(2,1,0)(2,0,0)[12] with drift : -333.7869
## ARIMA(3,1,1)(2,0,0)[12] with drift : -334.3744
## ARIMA(1,1,0)(2,0,0)[12] with drift : -328.5463
## ARIMA(3,1,0)(2,0,0)[12] with drift : -334.4836
## ARIMA(3,1,2)(2,0,0)[12] with drift : -332.688
## ARIMA(2,1,1)(2,0,0)[12] : -334.2453
##
## Now re-fitting the best model(s) without approximations...
##
## ARIMA(2,1,1)(2,0,0)[12] with drift : -315.7639
##
## Best model: ARIMA(2,1,1)(2,0,0)[12] with drift
## Series: UR_ts
## ARIMA(2,1,1)(2,0,0)[12] with drift
##
## Coefficients:
## ar1 ar2 ma1 sar1 sar2 drift
## 0.2212 0.4760 0.3741 0.0758 -0.1832 -0.0538
## s.e. 0.1853 0.1336 0.2015 0.0670 0.0692 0.0321
##
## sigma^2 estimated as 0.0154: log likelihood=165.12
## AIC=-316.24 AICc=-315.76 BIC=-291.76
##
## Training set error measures:
## ME RMSE MAE MPE MAPE MASE
## Training set -0.001081206 0.1222982 0.09234395 0.01085794 1.094762 0.06791711
## ACF1
## Training set -0.002100423
## Series: UR_ts
## ARIMA(2,1,1)(2,0,0)[12] with drift
##
## Coefficients:
## ar1 ar2 ma1 sar1 sar2 drift
## 0.2212 0.4760 0.3741 0.0758 -0.1832 -0.0538
## s.e. 0.1853 0.1336 0.2015 0.0670 0.0692 0.0321
##
## sigma^2 estimated as 0.0154: log likelihood=165.12
## AIC=-316.24 AICc=-315.76 BIC=-291.76
##
## Training set error measures:
## ME RMSE MAE MPE MAPE MASE
## Training set -0.001081206 0.1222982 0.09234395 0.01085794 1.094762 0.06791711
## ACF1
## Training set -0.002100423
##
## Ljung-Box test
##
## data: Residuals from ARIMA(2,1,1)(2,0,0)[12] with drift
## Q* = 22.909, df = 18, p-value = 0.1941
##
## Model df: 6. Total lags used: 24
##
## Ljung-Box test
##
## data: Residuals from ARIMA(2,1,1)(2,0,0)[12] with drift
## Q* = 22.909, df = 18, p-value = 0.1941
##
## Model df: 6. Total lags used: 24
##
## Ljung-Box test
##
## data: Residuals from ARIMA(2,1,1)(2,0,0)[12] with drift
## Q* = 22.909, df = 18, p-value = 0.1941
##
## Model df: 6. Total lags used: 24
##
## Ljung-Box test
##
## data: Residuals from ARIMA(2,1,1)(2,0,0)[12] with drift
## Q* = 22.909, df = 18, p-value = 0.1941
##
## Model df: 6. Total lags used: 24
## Point Forecast Lo 80 Hi 80 Lo 95 Hi 95
## Jun 2017 3.081191 2.92217160 3.240211 2.83799167 3.324391
## Jul 2017 3.038107 2.73870440 3.337510 2.58021017 3.496004
## Aug 2017 3.002284 2.54146274 3.463106 2.29751853 3.707050
## Sep 2017 2.958886 2.33757863 3.580193 2.00867838 3.909094
## Oct 2017 2.903089 2.11954185 3.686637 1.70475679 4.101422
## Nov 2017 2.857670 1.91591476 3.799425 1.41737980 4.297960
## Dec 2017 2.818959 1.72262214 3.915295 1.14225664 4.495661
## Jan 2018 2.761562 1.51592343 4.007200 0.85652230 4.666601
## Feb 2018 2.677773 1.28794328 4.067602 0.55221196 4.803334
## Mar 2018 2.619598 1.09103501 4.148160 0.28186278 4.957333
## Apr 2018 2.561125 0.89906044 4.223189 0.01921686 5.103032
## May 2018 2.484131 0.69370241 4.274559 -0.25409301 5.222354
## Jun 2018 2.387205 0.46898271 4.305428 -0.54646299 5.320874
## Jul 2018 2.306623 0.26322945 4.350017 -0.81847796 5.431725
## Aug 2018 2.244792 0.07876348 4.410822 -1.06786302 5.557448
## Sep 2018 2.200616 -0.08484597 4.486078 -1.29469640 5.695928
## Oct 2018 2.137103 -0.26465382 4.538860 -1.53606716 5.810273
## Nov 2018 2.092636 -0.42213477 4.607407 -1.75337399 5.938646
## Dec 2018 2.030307 -0.59430404 4.654919 -1.98368945 6.044304
## Jan 2019 1.966523 -0.76480820 4.697854 -2.21068767 6.143734
## Feb 2019 1.919026 -0.91604167 4.754094 -2.41683586 6.254888
## Mar 2019 1.855124 -1.08081675 4.791065 -2.63500987 6.345258
## Apr 2019 1.791178 -1.24291810 4.825274 -2.84907154 6.431427
## May 2019 1.725811 -1.40385907 4.855482 -3.06060652 6.512229
## Point Forecast Lo 80 Hi 80 Lo 95 Hi 95
## Jun 2017 3.081191 2.92217160 3.240211 2.83799167 3.324391
## Jul 2017 3.038107 2.73870440 3.337510 2.58021017 3.496004
## Aug 2017 3.002284 2.54146274 3.463106 2.29751853 3.707050
## Sep 2017 2.958886 2.33757863 3.580193 2.00867838 3.909094
## Oct 2017 2.903089 2.11954185 3.686637 1.70475679 4.101422
## Nov 2017 2.857670 1.91591476 3.799425 1.41737980 4.297960
## Dec 2017 2.818959 1.72262214 3.915295 1.14225664 4.495661
## Jan 2018 2.761562 1.51592343 4.007200 0.85652230 4.666601
## Feb 2018 2.677773 1.28794328 4.067602 0.55221196 4.803334
## Mar 2018 2.619598 1.09103501 4.148160 0.28186278 4.957333
## Apr 2018 2.561125 0.89906044 4.223189 0.01921686 5.103032
## May 2018 2.484131 0.69370241 4.274559 -0.25409301 5.222354
## Jun 2018 2.387205 0.46898271 4.305428 -0.54646299 5.320874
## Jul 2018 2.306623 0.26322945 4.350017 -0.81847796 5.431725
## Aug 2018 2.244792 0.07876348 4.410822 -1.06786302 5.557448
## Sep 2018 2.200616 -0.08484597 4.486078 -1.29469640 5.695928
## Oct 2018 2.137103 -0.26465382 4.538860 -1.53606716 5.810273
## Nov 2018 2.092636 -0.42213477 4.607407 -1.75337399 5.938646
## Dec 2018 2.030307 -0.59430404 4.654919 -1.98368945 6.044304
## Jan 2019 1.966523 -0.76480820 4.697854 -2.21068767 6.143734
## Feb 2019 1.919026 -0.91604167 4.754094 -2.41683586 6.254888
## Mar 2019 1.855124 -1.08081675 4.791065 -2.63500987 6.345258
## Apr 2019 1.791178 -1.24291810 4.825274 -2.84907154 6.431427
## May 2019 1.725811 -1.40385907 4.855482 -3.06060652 6.512229
We can see that the residuals are normally distributed and we can forecast. By both models unemployment will go down and it is in line with our expectations.